The purpose of creating a suitability model is to identify the best locations for some phenomenon. Phenomena for which a suitability model a be constructed for are:

 

• A new shopping center
• Habitat for conservation
• A secure military outpost

 

When looking at a map displaying suitability at each location, it can be difficult to identify which areas are the very best. Generally, the most suitable locations (locations with the highest values) are not contiguous and potentially isolated.

 

By incorporating the Locate Regions tool into a suitability model, one can avoid arbitrarily delineating supposed locations of the most suitable areas

In many cases, you will have knowledge of the size, number of regions necessary, and the spatial relationships of the regions that you are looking for. These characteristics , and others, can be tailored to suit the requirements for one, or multiple, regions.

 

The Locate Regions tool allows the following parameters to be controlled:

 

• Total area of the region(s)
• Number of regions
• Shape (type and orientation)
• Trade-off between value (utility) and shape
• Minimum and maximum size of each region
• Minimum and maximum distance between regions
• Inclusion of existing regions

 

Controlling the above parameters will generate the best contiguous regions of a specified size meeting the desired spatial constraints within the study area.

Depending on a given scenario, the total amount of area grouped into a single region varies. For example:

 

• A new shopping mall needs to be 200,000 square feet.
• A conservation organization has funds to purchase 1000 square kilometers.
• Military engineers need 15 acres for a missile site.

 

Entering in a value and the associated units for the Total_Area parameter will result in a single feature of the specified area.

 

Click on the following scenarios to explore the effects of changing the Total_Area value. Clicking the "BACK" button that appears at the top of the screen allows for comparison to the original region.

 

• Large total area
• Small total area

 

The associated units drop-down list is disabled until a value is entered but the default changes to the units of the input raster when the input raster is entered. If the input raster is in meters, the units default to hectares; if feet the units default to acres.

In cases where cells need to be grouped into a specified number of regions of equal area, the Locate Regions tool requires values for total area and number of regions.

 

If one region is specified, it will span the required total area. If the number of regions is greater than one, then the total area will be divided among each region.

 

Click on the following scenarios to explore the effects of changing the number of regions.

 

• One region
• Two regions
• Four regions

Certain instances may require control over the shape and orientation of a region. For example:

 

• Shopping centers are often square or rectangular
• Airports often face their runways into the prevailing winds
• Rock quarries are aligned parallel to bedrock formations

 

The Locate Regions tool provides multiple options regarding shape and orientation.

 

Shape:

• Circle
• Ellipse
• Equillateral triangle
• Square
• Pentagon
• Hexagon
• Octagon

 

Orientation:

• North-south
• Northeast-southwest
• East-west
• Southeast-northwest

Due to the disconnected nature of cells with the highest suitability values in most suitability surfaces, a trade-off needs to be made between shape and value when outlining a region.

 

The Shape_Tradeoff parameter allows for control over the relative importance, or weighting, of shape and value, or utility. This parameter identifies the weighting for the cells when growing the potential candidate regions in the algorithm behind the tool. The weighting is a trade-off between a cell’s contribution for maintaining the shape relative to the utility contribution of each cell’s attribute value.

 

The higher the value indicates maintaining the shape of the region is more important than

selecting higher attribute values.

 

Click on the following instances to see how changing the trade-off value impacts the resulting region

 

• Emphasis placed on shape
• Emphasis placed on utility

The previous scenarios have assumed equal area for each region, but other plans may be more flexible on the size of any given region. Therefore, minimum and maximum areas of regions can be specified.

 

• Certain housing units may require a minimum area only
• An landscape architect is given a maximum amount of area to work with
• A new shopping mall needs enough space for 15 stores and maximum area is limited by local zoning laws

 

 

When more than one region is specified, no region can exceed the maximum specified area.

 

Click on the scenarios below to investigate how minimum and maximum area constraints effect regions:

 

• Minimum area value only
• Maximum area value only
• Minimum and maximum area values specified
• Wide range between minimum and maximum area values
• Small range between minimum and maximum area values

In addition to controlling characteristics pertaining to the regions themselves, other parameters within the Locate Regions tool dictate how the regions are arranged throughout the study area. Two parameters, Minimum_Distance and Maximum_Distance, are useful for specifying a minimum or maximum distance between any two regions.

 

By indicating a value for minimum distance, no two regions can be within that distance. Similarly, no two regions can be farther apart than a specified maximum distance.

 

The maximum distance includes distances from existing regions but distances from excluded areas are irrelevant.

 

Click on the options below to explore differences in the minimum and maximum distance values.

 

• Minimum distance value only
• Maximum distance value only
• Minimum and maximum distance values specified
• Wide range between minimum and maximum distance values
• Small range between minimum and maximum distance values

In some cases, a suitability model may be built to locate new regions, when some locations potentially already exist. Existing locations may need to be taken into consideration when planning the areas for new regions. For example:

 

• Two fire stations already serve half of a city, and a new fire station should serve the uncovered areas
• Several wildlife refuges need to be included in plans for a state-wide wildlife connectivity initiative

 

The input can be a raster or feature data set. Any location in the raster with a valid value is considered as already allocated.

 

Click below to investigate how different existing regions can alter the spatial arrangement of new regions.

 

Number of existing regions:

• One
• Two

 

Output regions with:

• No existing regions
• One existing region
• Two existing regions

Locating regions is a four step process:

 

1. Before the tool is run, all locations (cells) that should not be considered in the location process (e.g. bodies of water, slopes too steep, existing buildings) are removed.
2. Parameters defining the characteristics of the desired region(s) – e.g. size, shape, and region orientation are specified.
3. Most likely candidate regions from the input raster are identified based on the desired region characteristics .
4. “Best” region(s) from the candidate regions are selected using a user-defined evaluation criterion – e.g., region(s) with the highest average value

 

An algorithm called parameterized region-growth generates and expands candidate regions from individual cells, or seeds. The specified trade-off value dictates which contiguous cells are added to a region. Potential candidate regions will continue to grow until the specified area requirements for the region are met.

 

To select the “best” region(s) each candidate region identified above is evaluated based on the Region Evaluation method (e.g. Highest average value, Highest sum, Highest median value, etc.) and the inter-region evaluation criterion (minimum and maximum distances among regions) to produce the best configuration of regions.

 

At this stage of the workflow, the “best” locations for a phenomenon have been identified. Results should be analyzed to gain further insight into decision-making.

 

Acknowledgements

We thank Steven Lamonde of Johnson State College and the Vermont Center for Geographic Information for their contributions.